Conferencias UTE - FCII, INCISCOS 2016

Por defecto: 
Automatic identification of a playing card through kNN using a Raspberry Pi 3
Juan Daniel Estévez, Holger Ortega, Rodrigo Tufiño

Última modificación: 2017-02-16


This academic work has the main objective to develop a system for automatic recognition of an English playing card located on a table using computer vision techniques for capturing, preprocessing, and segmenting the image independently of the orientation and the depression angle. The algorithm used as a classifier to recognize the card is k-nearest neighbor (kNN). At training stage, a set based on a list of alphanumeric characters was used. The result of the classification was sent to an audio output using a converter from text to voice. This algorithm was implemented in an embedded system Raspberry Pi 3 under the operative system Raspbian Jessie. The system developed has an accuracy of 95% and an average wait- response of 5 seconds taking into account the audio playing.

Texto completo: PDF